Table of Contents
💡 Note: AI-OPS is an active research prototype.
Overview
AI-OPS is a Penetration Testing AI Assistant that leverages open-source Large Language Models (LLMs) to explore the role of Generative AI in ethical hacking. With a focus on accessibility and practical use, it aims to accelerate common tasks in pentesting such as exploit development, vulnerability research, and code analysis.
Disclaimer: AI-OPS goal is to support human operators rather than automate penetration testing activities entirely,
ensuring that AI remains a supplementary tool during real-world workflows. As any other automation tool, it doesn't
replace operator competence, neither knowledge: AI won't do the work for you, but it may help in the process.
Key Features
- 🚀 Full Open-Source: There is no reliance on third-party LLM providers; use any model you prefer with Ollama.
- 🔍 Web Search: The AI assistant delivers up-to-date responses by performing online searches via Google.
Supported Models
AI-OPS supports virtually any LLM that can be hosted with Ollama, allowing you to tailor the assistant to your available resources; for detailed setup instructions, see Configuration. While flexibility is key, note that performance may vary depending on the model used; below is a list of currently implemented models.
| Name | Notes |
|---|---|
| DeepSeek-R1 | New integration. |
| Mistral 7B | Using non quantized mistral mistral:7b-instruct-v0.3-q8_0 gives better results in exploit tasks. |
| Gemma 2 | Better compared to mistral in vulnerability research tasks. |
Note: AI-OPS prioritizes smaller LLMs to ensure accessibility and optimal performance, even on limited hardware.
Quickstart
Requirements
To get started with AI-OPS, ensure you have the following dependencies installed:
- Python (>= 3.11): for AI-OPS CLI interface.
- Ollama (>= 0.3.0): for LLM inference.
- Docker : for AI-OPS API.
Get Started
Start by cloning the repository:
git clone https://github.com/antoninoLorenzo/AI-OPS.git
cd AI-OPS
Then configure Ollama, you can refer to their documentation for additional details:
ollama run MODEL
💡 Tip: If you lack mid/high-end GPUs to run LLMs locally you can follow my guide on how to run Ollama on Google Colab.
Build and run the Docker container for the AI-OPS API using the following command. Replace ENDPOINT with the URL of your Ollama instance and MODEL with the name of the model you wish to use (e.g., Mistral 7B):
docker build -t ai-ops:api-dev --build-arg ollama_endpoint=ENDPOINT --build-arg ollama_model=MODEL .
docker run -p 8000:8000 ai-ops:api-dev
To start interacting with AI-OPS, install and run the ai-ops-cli command-line client and replace AI-OPS_API_ADDRESS with the address of your running Docker container:
pip install .
ai-ops-cli --api AI-OPS_API_ADDRESS
Additional Information
User Documentation
⚖️ Ethical and Legal Considerations
AI-OPS is designed as a penetration testing tool intended for academic and educational purposes only. Its primary goal is to assist cybersecurity professionals and enthusiasts in enhancing their understanding and skills in penetration testing through the use of AI-driven automation and tools.
Disclaimer
The creators and contributors of AI-OPS are not responsible for any misuse of this tool. By using AI-OPS, you agree to take full responsibility for your actions and to use the tool in a manner that is ethical, legal, and in accordance with the intended purpose.
This project is provided "as-is" without any warranties, express or implied. The creators are not liable for any damages or legal repercussions resulting from the use of this tool.